A REPEATING PATTERN OF ERROR?
22 July 2005 - Safe Speed

Following a freedom of information request the Department for Transport has revealed that the authors of a new report into speed camera effectiveness are the same as last year.

But the DfT has just called a halt to speed camera expansion apparently because they were not satisfied by last years'' report.

Safe Speed wonders how they can have confidence in this years'' report.

Paul Smith, founder of the Safe Speed road safety campaign
(www.safespeed.org.uk) said: ''Last year''s report was 115 pages, but I knew that the content didn''t justify the conclusions after just 15 minutes study. I made every effort to inform people of the fact and after a long struggle involving ''Questions in the House'' it appears that the primary error source is now being admitted.''

That primary error is known as ''Regression To The Mean (RTTM)'', and is a pitfall extremely well known to statisticians. Consider the following sample crash statistics:

Year KSI accidents

1997 1
1998 1
1999 1
2000 4
2001 1 camera installed
2002 1
2003 1
2004 1

The rules for camera placement require 4 crashes in a three year period. In 1999 we have only had three crashes so no camera is allowed. In 2000 we have a bad year and when we get the year-end figures we can see that a camera can be justified. So in it goes. The site then returns to its long term norm of 1 crash per year but the figures for the camera look like this:

* 3 years before: 6 crashes
* 3 years after: 3 crashes
* ''benefit'': crashes cut by 50%

What''s really happened here is that a random grouping of crashes triggered the installation of the camera. The camera delivered no benefit but there was an illusion of benefit when the number of crashes at the site returned to the long term average.

''The trouble with RTTM is that you can get exactly the same effect by placing a garden Gnome at the site.'' explains Paul, ''Every motorist needs to understand how the DfT used figures with an exaggerated benefit to misinform the public. Every motorist needs to understand the regression to the mean trick. And we all need to know how it was possible for a Professor of statistics to make such a blunder. I REALLY want to know why they are using the same authors for this year''s report.''

www.safespeed.org.uk

More News
  For July 2005
  From Safe Speed
  For Associations
  Biker247.com Home Page